利用新闻情感数据炒股 python程序

清华大佬耗费三个月吐血整理的几百G的资源,免费分享!....>>>

from CAL.PyCAL import Date

start = '2010-01-01'                       # 回测起始时间
end = '2015-05-05'                         # 回测结束时间
benchmark = 'HS300'                        # 策略参考标准
universe = set_universe('HS300')               # 证券池,支持股票和基金
capital_base = 1000000                      # 起始资金
longest_history = 0                        # handle_data 函数中可以使用的历史数据最长窗口长度
refresh_rate = 1                          # 调仓频率,即每 refresh_rate 个交易日执行一次 handle_data() 函数
longest_history = 1


def initialize(account):                   # 初始化虚拟账户状态
    account.isBuyPeriod = False
    account.dayCount = 0

def handle_data(account):                  # 每个交易日的买入卖出指令
    account.dayCount += 1 
    if account.isBuyPeriod:                # 每60个工作日(3个月)调仓
        hist = account.get_history(longest_history)
        endDate = Date.fromDateTime(account.current_date)
        startDate = endDate - 30
        res =  DataAPI.NewsSentimentIndexGet(secID=account.universe, field=['secID', 'newsPublishDate', 'sentimentIndex'], beginDate=startDate.strftime('%Y%m%d'),endDate=endDate.strftime('%Y%m%d'))
        res = res.groupby('secID')
        
        # top 10%
        top10 = res.mean().sort('sentimentIndex', ascending=False).head(int(0.1*len(res)))
        buyList = list(top10.index)
        print u"%s 买入 : %s" % (endDate, buyList)
        
        # 等权重买入
        if len(buyList) != 0:
            singleCash = account.cash / len(buyList)
            for stock in buyList:
                approximationAmount = int(singleCash / hist[stock]['closePrice'][-1]/100.0) * 100
                order(stock, approximationAmount)

        account.isBuyPeriod = False
        account.dayCount = 0
    elif account.dayCount == 59:          # 调仓日前一日清空当前仓位
        for stock in account.valid_secpos:
            order_to(stock,0)
        account.isBuyPeriod = True